Deepfakes detection increasingly vital to law enforcement, fraud defense alike

Law enforcement agencies around the world are grappling with deepfakes in several different areas of investigation. A post on Reality Defender’s website explains how they are doing so, and aligns with comments from executives with iProov and Vida Group about the fraud environment for businesses in Asia.

Fortunately, more tools to protect businesses from deepfake-powered fraud are joining the arsenal of available defenses.

Identy.io has built a deepfake detection layer into its mobile face biometrics capture software. The new capability includes visual and temporal analysis to identify AI-generated content, but also injection attack detection (IAD) and prevention, along with Identy’s passive presentation attack detection (PAD).

Avast has launched Deepfake Guard software for Windows computers to detect scam videos on popular platforms. The software is included in the Gen Digital subsidiary’s Avast Premium Security and works with major platforms including Facebook, DailyMotion, Instagram, TikTok, Twitch, Vimeo, X, and YouTube, according to the announcement.

Criminal evidence to political misinformation

Reality Defender starts with the point that trust in judicial institutions depends on digital media authentication as claims of “AI-faked evidence” proliferate.

The post on deepfake detection in investigations makes the case that provenance-based methods like metadata analysis have been rendered obsolete and explains how deepfake detection is integrated with existing forensic workflows.

Agencies around the world are currently confronting deepfakes in child sexual assault material (CSAM), extortion and exploitation, financial crime and impersonation, disinformation and public safety threats and building AI-ready forensics, according to the post.

On disinformation threats, Reality Defender refers to a partnership it has with NATO StratCom. Under that partnership the company identified likely deepfake videos spreading misinformation about a political leader in a NATO country in 2022.

In a demonstration of the possible interplay between agentic AI and deepfakes, CEO Ben Colman shared the results of an experiment in which he fed an undisclosed AI agent a video of himself and a collection of celebrity photographs culled from Google Images on LinkedIn.

Six minutes of agentic AI effort and four minutes of human video editing later, a collection of celebrities can be seen replicating his movements, with no obvious visual tells that they are fake. The speed of production makes crafting, iterating and perfecting deepfake attacks much more realistic, and underlines that the problem has now gone beyond the fidelity of fakes.

“We are moving from ‘craft’ deepfakes to ‘assembly line’ deception,” Colman says in summary.

More attacks, more professional

iProov CTO Dominic Forrest tells BusinessWorld that the scale of synthetic identity fraud is widely underestimated, and will become better understood this year.

As that understanding dawns, he warns that businesses in the Philippines must be prepared to step up their prevention efforts, such as with quarterly or monthly biometric checks. Face biometrics used in onboarding must also include robust liveness detection.

Without these measures, criminals can create fake accounts or, in remote work situations, commit employment fraud to set up further attacks.

iProov’s biometrics and liveness detection customers in the Philippines include UnionBank.

Niki Luhur, founder and CEO of Indonesia-based Vida Group, says fraud in Southeast Asia has passed a tipping point, attracting organized crime and commercializing attacks enabled by generative AI.

Luhur addressed the audience at CISO Malaysia 2026 last Thursday, TNGlobal reports, warning that “When crime operates like an industry, security cannot rely on assumptions built for a different era.”

The layered approach Luhur recommends includes device integrity checks, biometric liveness detection and PKI authentication, unified in a cohesive identity verification system.

Article Topics

AI fraud  |  Avast  |  biometric liveness detection  |  deepfake detection  |  deepfakes  |  generative AI  |  iProov  |  Reality Defender  |  synthetic identity fraud  |  VIDA